Active Learning of Uniformly Accurate Inter-atomic Potentials for Materials Simulation
This work addresses the challenge of efficiently developing reliable inter-atomic potentials for materials science, representing an incremental improvement in active learning methods for this domain.
The authors tackled the problem of constructing accurate and transferable machine learning models for materials simulation by proposing an active learning procedure called DP-GEN, which produced uniformly accurate potential energy surface models for Al, Mg, and Al-Mg alloys with a minimal number of reference data.
An active learning procedure called Deep Potential Generator (DP-GEN) is proposed for the construction of accurate and transferable machine learning-based models of the potential energy surface (PES) for the molecular modeling of materials. This procedure consists of three main components: exploration, generation of accurate reference data, and training. Application to the sample systems of Al, Mg and Al-Mg alloys demonstrates that DP-GEN can produce uniformly accurate PES models with a minimal number of reference data.